Article Summary: “Pictures into Words” by Brian Stewart

“Pictures into Words” by Brian Stewart describes a study comparing user indexing, i.e. crowdsourced tagging, with professional indexing. The article first summarizes Panofsky’s and Shatford’s methods for classifying images. Panofsky outlined three levels of subject matter/meaning:
  1. Pre-iconographical description (objects or events) 
  2. Iconographical analysis (themes)
  3. Iconographical interpretation (meaning) 
Shatford emphasized two dimensions of images, ‘of’ and ‘about’ (sometimes called an image’s ofness and aboutness.) Shatford also defined four facets of an image: who, what, where, and when.


Stewart argues that indexers have traditionally taken a positivist approach by labeling mostly specific, objective subjects. Today, however, users are in greater need of abstract subjects (interpretive indexing). The article gives the example of abstract subjects like “happy” or “peace.”


Part I of the study examined 11 professional indexers. Stewart broke down the subject terms they chose into specific, generic, and abstract categories. The professionals’ chosen terms for the set of photographs was 52% specific, 46.2% generic, and 1.8% abstract. They were reluctant to use abstract terms for fear of being too general. Professional indexers refrained from selecting abstract terms even when those terms would be useful, and Stewart argues that they made false assumptions about what the user needs.
Part II of the study examined social tagging, i.e. crowdsourcing on Flickr. 66 taggers tagged 33 photos, 11 of which were untitled and 22 of which were titled and given a photographer name. Interestingly, whether a photo was titled or not had little influence on taggers’ subject selections. The user group showed a strong preference for generic terms (60.3%) and about equal selection of specific and abstract terms (19.62% and 20%, respectively). Taggers were then given a “training intervention” which taught them the Panofsky/Shatford matrix and encouraged increased attention to abstract terms. After training, the users selected more than double the number of terms as before and the number of abstract terms multiplied by 5.1.


Stewart believes that users know better than professionals what they are searching for and what tags are most appropriate. Users are more likely to select abstract terms, especially after special training. Crowdsourcing the work of indexing actively involves users in the process of discovery. Practically speaking, professionals may become dependent on crowdsourcing the work of indexing as the volume of digital materials continues to mushroom. The greatest challenge is, as Stewart addresses in this study, keeping users motivated and active. Stewart found that tagging activity from users stagnated after ten days. For crowdsourcing projects to be sustainable, users need to feel invested in the project and like they are part of a community. It is the professional’s task to mentor and foster this community.


Stewart, Brian. “Pictures into Words.” The Indexer vol. 33, no. 1 (March 2015): 8-25.

Comments

  1. StevenMarch 1, 2018 at 1:05 PM

    Good work! This is a great article explaining the range of approaches to data entry, including crowdsourcing. It's interesting that he notes how "addicted" professionals can become to the crowds helping to organize our digital abundance!

    Dr. MacCall

    ReplyDelete

Post a Comment

Popular posts from this blog

Article Summary: "The Importance of Interoperability: Lessons from the Digital Public Library of America"

Digital libraries like the DPLA , Europeana , the World Digital Library , and the National Library of France's Gallica (just to name a few) are trying to make cultural heritage not only accessible to those looking for it, but also something one could stumble upon during a web search. Metadata gives institutions the ability to make the DPLA's record of a 1983 interview with Elie Wiesel just as discoverable as his interview with Oprah on YouTube.  Don't get me wrong - I love Oprah and I love YouTube. But YouTube videos aren't made and published with longevity in mind. The DPLA fosters a better balance between preservation and access - it gives PR to smaller cultural heritage institutions and their collections across the country.  The DPLA increases the number of "hits" institutions receive on their websites by means of good, interoperable metadata.  Interoperability - "The ability of multiple systems with different hardware and software platforms...
Read more

Article Summary: “When Archivists and Digital Asset Managers Collide: Tensions and Ways Forward" by Anthony Cocciolo

    In the digital era, some titles in library science have been rebranded. An association with data and computing is expected to attract prestige and money. LIS programs are becoming iSchools, and LIS graduates are more frequently pursuing jobs like “data architect,” “digital librarian,” or “digital preservationist.” Sometimes these positions are rebranded versions of the same thing, but sometimes they are truly new jobs. Cocciolo’s article discusses the “digital asset manager” and how this position collides with that of the archivist. He carries out a case study at an art museum with one digital asset manager and two archivists. The digital asset manager was brought on in 2012 to manage the digital asset management (DAM) software. At first, the digital asset manager worked solely with the photography department, since DAM is most established in digital photography. Tensions arose, however, when the DAM spread to other departments throughout the museum. Essential...
Read more